Abstract
<p>The Adolescent Psychotic-like Symptom Screener (APSS) is widely used to screen adolescents for psychotic experiences (PEs), but there are few rigorous psychometric analyses of the instrument. In this study, we apply Item Response Theory (IRT) to APSS responses from a large sample of middle and high school students in Massachusetts (N = 37,359, 49% male, mean age 14.9 years). We compare polytomous IRT models, evaluate item and test functioning, test for differential item functioning (DIF), and examine demographic predictors of PEs. We find that the data are consistent with unidimensionality and a graded response model fits the data well. Reliability is maximized about 1.5 standard deviations above the mean and is poor at low trait levels. We identify two practically significant DIF terms by sex, but they operate in opposite directions and are unlikely to generate large biases in overall scores. Latent regression analyses show that older students and male students report slightly lower levels of PEs, with a modest sex-by-age interaction. Our results support the use of the APSS as an efficient screener for identifying adolescents with elevated PEs who may benefit from further assessment, while cautioning that it is not well suited to distinguish participants with low PEs.</p>